def do_table_results(self, parameter_name, parameter_values, scores):
rows = [['parameter_value_' + parameter_name, 'score']]
x = PrettyTable()
x.field_names = ['parameter_value_' + parameter_name, 'score']
for index, parameter_value in enumerate(parameter_values):
rows.append([parameter_value, scores[index]])
x.add_row([parameter_value, scores[index]])
if self.output_path is not None:
save_csv(
self.output_path + '/table_scores_parameter_' + parameter_name,
rows)
print(x)
def class_frequencies_separate(config, output_path, values, labels, output_labels, visualize, dataframe, verbose):
unique, counts = np.unique(output_labels, return_counts=True)
rows = [['class', 'predicted_samples_distribution']]
x = PrettyTable()
x.field_names = ['class', 'predicted_samples_distribution']
for index, _class in enumerate(unique):
number_samples = counts[index]
rows.append([_class, number_samples])
x.add_row([_class, number_samples])
if output_path is not None:
save_csv(output_path + '/samples_distribution', rows)
print(x)
def show_metrics_table(config, output_path, values, labels, output_labels, visualize, dataframe, verbose):
rows = [[''] + config['metrics']]
x = PrettyTable()
x.field_names = [''] + config['metrics']
scores = []
for metric in config['metrics']:
scores.append(run_evaluation_metric(metric, values, labels, output_labels))
rows.append(['result'] + scores)
x.add_row(['result'] + scores)
if output_path is not None:
save_csv(output_path + '/table_scores', rows)
print(x)
def class_frequencies(config, output_path, values, labels, output_labels, visualize, dataframe, verbose):
unique1, counts1 = np.unique(labels, return_counts=True)
#counts1, unique1 = zip(*sorted(zip(counts1, unique1), reverse=True)) -> show sorted
unique2, counts2 = np.unique(output_labels, return_counts=True)
rows = [['class', 'original_samples_distribution', 'predicted_samples_distribution']]
x = PrettyTable()
x.field_names = ['class', 'original_samples_distribution', 'predicted_samples_distribution']
for index, _class in enumerate(unique1):
number_samples1 = counts1[index]
number_samples2 = counts2[np.where(unique2 == _class)[0]][0]
rows.append([_class, number_samples1, number_samples2])
x.add_row([_class, number_samples1, number_samples2])
if output_path is not None:
save_csv(output_path + '/samples_distribution', rows)
print(x)